A network of N elements is studied in terms of a deterministic globall
y coupled map which can be chaotic. There exists a range of values for
the parameters of the mag where the number of different macroscopic c
onfigurations N(N) is very large, N(N) similar to exp root c(a)N, and
there is violation of self-averaging. The time averages of functions,
which depend on a single element, computed over a time T, have probabi
lity distributions that for any N do not collapse to a delta function,
for increasing T. This happens for both chaotic and regular motion, i
.e., positive or negative Lyapunov exponent.